ruby-macho 4.1.0 → 6.0.0

data/lib/macho/exceptions.rb

@@ -37,7 +37,7 @@ module MachO
   # Raised when a file is too short to be a valid Mach-O file.
   class TruncatedFileError < NotAMachOError
     def initialize
-      super "File is too short to be a valid Mach-O"
+      super("File is too short to be a valid Mach-O")
     end
   end
 
@@ -45,21 +45,21 @@ module MachO
   class MagicError < NotAMachOError
     # @param num [Integer] the unknown number
     def initialize(magic)
-      super "Unrecognized Mach-O magic: 0x%02<magic>x" % { :magic => magic }
+      super("Unrecognized Mach-O magic: 0x%02<magic>x" % { :magic => magic })
     end
   end
 
   # Raised when a file is a Java classfile instead of a fat Mach-O.
   class JavaClassFileError < NotAMachOError
     def initialize
-      super "File is a Java class file"
+      super("File is a Java class file")
     end
   end
 
   # Raised when a a fat Mach-O file has zero architectures
   class ZeroArchitectureError < NotAMachOError
     def initialize
-      super "Fat file has zero internal architectures"
+      super("Fat file has zero internal architectures")
     end
   end
 
@@ -71,24 +71,24 @@ module MachO
       # @param cpusubtype_fat [Integer] the CPU subtype in the fat header
       # @param cputype_macho [Integer] the CPU type in the macho header
       # @param cpusubtype_macho [Integer] the CPU subtype in the macho header
-      super ("Mismatch between cputypes >> 0x%08<fat_cputype>x and 0x%08<macho_cputype>x\n" \
-             "and/or cpusubtypes >> 0x%08<fat_cpusubtype>x and 0x%08<macho_cpusubtype>x" %
-            { :fat_cputype => fat_cputype, :macho_cputype => macho_cputype,
-              :fat_cpusubtype => fat_cpusubtype, :macho_cpusubtype => macho_cpusubtype })
+      super("Mismatch between cputypes >> 0x%08<fat_cputype>x and 0x%08<macho_cputype>x\n" \
+            "and/or cpusubtypes >> 0x%08<fat_cpusubtype>x and 0x%08<macho_cpusubtype>x" %
+        { :fat_cputype => fat_cputype, :macho_cputype => macho_cputype,
+          :fat_cpusubtype => fat_cpusubtype, :macho_cpusubtype => macho_cpusubtype })
     end
   end
 
   # Raised when a fat binary is loaded with MachOFile.
   class FatBinaryError < MachOError
     def initialize
-      super "Fat binaries must be loaded with MachO::FatFile"
+      super("Fat binaries must be loaded with MachO::FatFile")
     end
   end
 
   # Raised when a Mach-O is loaded with FatFile.
   class MachOBinaryError < MachOError
     def initialize
-      super "Normal binaries must be loaded with MachO::MachOFile"
+      super("Normal binaries must be loaded with MachO::MachOFile")
     end
   end
 
@@ -96,7 +96,7 @@ module MachO
   class CPUTypeError < MachOError
     # @param cputype [Integer] the unknown CPU type
     def initialize(cputype)
-      super "Unrecognized CPU type: 0x%08<cputype>x" % { :cputype => cputype }
+      super("Unrecognized CPU type: 0x%08<cputype>x" % { :cputype => cputype })
     end
   end
 
@@ -105,8 +105,8 @@ module MachO
     # @param cputype [Integer] the CPU type of the unknown pair
     # @param cpusubtype [Integer] the CPU sub-type of the unknown pair
     def initialize(cputype, cpusubtype)
-      super "Unrecognized CPU sub-type: 0x%08<cpusubtype>x " \
-            "(for CPU type: 0x%08<cputype>x" % { :cputype => cputype, :cpusubtype => cpusubtype }
+      super("Unrecognized CPU sub-type: 0x%08<cpusubtype>x " \
+            "(for CPU type: 0x%08<cputype>x" % { :cputype => cputype, :cpusubtype => cpusubtype })
     end
   end
 
@@ -114,7 +114,7 @@ module MachO
   class FiletypeError < MachOError
     # @param num [Integer] the unknown number
     def initialize(num)
-      super "Unrecognized Mach-O filetype code: 0x%02<num>x" % { :num => num }
+      super("Unrecognized Mach-O filetype code: 0x%02<num>x" % { :num => num })
     end
   end
 
@@ -122,7 +122,15 @@ module MachO
   class LoadCommandError < MachOError
     # @param num [Integer] the unknown number
     def initialize(num)
-      super "Unrecognized Mach-O load command: 0x%02<num>x" % { :num => num }
+      super("Unrecognized Mach-O load command: 0x%02<num>x" % { :num => num })
+    end
+  end
+
+  # Raised when a load command has an invalid size.
+  class LoadCommandSizeError < NotAMachOError
+    # @param size [Integer] the invalid size
+    def initialize(size)
+      super("Invalid Mach-O load command size: #{size}")
     end
   end
 
@@ -130,7 +138,7 @@ module MachO
   class LoadCommandNotCreatableError < MachOError
     # @param cmd_sym [Symbol] the uncreatable load command's symbol
     def initialize(cmd_sym)
-      super "Load commands of type #{cmd_sym} cannot be created manually"
+      super("Load commands of type #{cmd_sym} cannot be created manually")
     end
   end
 
@@ -141,8 +149,8 @@ module MachO
     # @param expected_arity [Integer] the number of arguments expected
     # @param actual_arity [Integer] the number of arguments received
     def initialize(cmd_sym, expected_arity, actual_arity)
-      super "Expected #{expected_arity} arguments for #{cmd_sym} creation, " \
-            "got #{actual_arity}"
+      super("Expected #{expected_arity} arguments for #{cmd_sym} creation, " \
+            "got #{actual_arity}")
     end
   end
 
@@ -150,7 +158,7 @@ module MachO
   class LoadCommandNotSerializableError < MachOError
     # @param cmd_sym [Symbol] the load command's symbol
     def initialize(cmd_sym)
-      super "Load commands of type #{cmd_sym} cannot be serialized"
+      super("Load commands of type #{cmd_sym} cannot be serialized")
     end
   end
 
@@ -158,8 +166,8 @@ module MachO
   class LCStrMalformedError < MachOError
     # @param lc [MachO::LoadCommand] the load command containing the string
     def initialize(lc)
-      super "Load command #{lc.type} at offset #{lc.view.offset} contains a " \
-            "malformed string"
+      super("Load command #{lc.type} at offset #{lc.view.offset} contains a " \
+            "malformed string")
     end
   end
 
@@ -167,7 +175,7 @@ module MachO
   class OffsetInsertionError < ModificationError
     # @param offset [Integer] the invalid offset
     def initialize(offset)
-      super "Insertion at offset #{offset} is not valid"
+      super("Insertion at offset #{offset} is not valid")
     end
   end
 
@@ -175,9 +183,9 @@ module MachO
   class HeaderPadError < ModificationError
     # @param filename [String] the filename
     def initialize(filename)
-      super "Updated load commands do not fit in the header of " \
+      super("Updated load commands do not fit in the header of " \
             "#{filename}. #{filename} needs to be relinked, possibly with " \
-            "-headerpad or -headerpad_max_install_names"
+            "-headerpad or -headerpad_max_install_names")
     end
   end
 
@@ -185,14 +193,14 @@ module MachO
   class DylibUnknownError < RecoverableModificationError
     # @param dylib [String] the unknown shared library name
     def initialize(dylib)
-      super "No such dylib name: #{dylib}"
+      super("No such dylib name: #{dylib}")
     end
   end
 
   # Raised when a dylib is missing an ID
   class DylibIdMissingError < RecoverableModificationError
     def initialize
-      super "Dylib is missing a dylib ID"
+      super("Dylib is missing a dylib ID")
     end
   end
 
@@ -200,7 +208,7 @@ module MachO
   class RpathUnknownError < RecoverableModificationError
     # @param path [String] the unknown runtime path
     def initialize(path)
-      super "No such runtime path: #{path}"
+      super("No such runtime path: #{path}")
     end
   end
 
@@ -208,7 +216,7 @@ module MachO
   class RpathExistsError < RecoverableModificationError
     # @param path [String] the extant path
     def initialize(path)
-      super "#{path} already exists"
+      super("#{path} already exists")
     end
   end
 
@@ -216,7 +224,7 @@ module MachO
   class UnimplementedError < MachOError
     # @param thing [String] the thing that is unimplemented
     def initialize(thing)
-      super "Unimplemented: #{thing}"
+      super("Unimplemented: #{thing}")
     end
   end
 
@@ -225,8 +233,8 @@ module MachO
   class FatArchOffsetOverflowError < MachOError
     # @param offset [Integer] the offending offset
     def initialize(offset)
-      super "Offset #{offset} exceeds the 32-bit width of a fat_arch offset. " \
-            "Consider merging with `fat64: true`"
+      super("Offset #{offset} exceeds the 32-bit width of a fat_arch offset. " \
+            "Consider merging with `fat64: true`")
     end
   end
 

data/lib/macho/fat_file.rb

@@ -96,7 +96,12 @@ module MachO
 
       @filename = filename
       @options = opts
-      @raw_data = File.binread(@filename)
+      File.open(@filename, "rb") do |file|
+        @raw_data = file.read(Headers::FatHeader.bytesize)
+        @raw_data ||= ""
+        populate_fat_header
+        @raw_data << file.read.to_s
+      end
       populate_fields
     end
 
@@ -165,7 +170,7 @@ module MachO
     # All load commands responsible for loading dylibs in the file's Mach-O's.
     # @return [Array<LoadCommands::DylibCommand>] an array of DylibCommands
     def dylib_load_commands
-      machos.map(&:dylib_load_commands).flatten
+      machos.flat_map(&:dylib_load_commands)
     end
 
     # Changes the file's dylib ID to `new_id`. If the file is not a dylib,
@@ -199,7 +204,7 @@ module MachO
       # Individual architectures in a fat binary can link to different subsets
       # of libraries, but at this point we want to have the full picture, i.e.
       # the union of all libraries used by all architectures.
-      machos.map(&:linked_dylibs).flatten.uniq
+      machos.flat_map(&:linked_dylibs).uniq
     end
 
     # Changes all dependent shared library install names from `old_name` to
@@ -229,7 +234,7 @@ module MachO
     # @see MachOFile#rpaths
     def rpaths
       # Can individual architectures have different runtime paths?
-      machos.map(&:rpaths).flatten.uniq
+      machos.flat_map(&:rpaths).uniq
     end
 
     # Change the runtime path `old_path` to `new_path` in the file's Mach-Os.
@@ -283,6 +288,16 @@ module MachO
       repopulate_raw_machos
     end
 
+    # Replaces every embedded signature with a pure-Ruby ad-hoc signature.
+    # @param identifier [String, nil] the signing identifier
+    # @return [void]
+    def codesign!(identifier: nil)
+      identifier ||= CodeSigning.identifier(canonical_macho, filename)
+      machos.each { |macho| macho.codesign!(:identifier => identifier) }
+      repopulate_resized_raw_machos
+      nil
+    end
+
     # Extract a Mach-O with the given CPU type from the file.
     # @example
     #  file.extract(:i386) # => MachO::MachOFile
@@ -402,6 +417,37 @@ module MachO
       end
     end
 
+    # Rebuild the fat header and slice layout after internal Mach-Os change size.
+    # {#repopulate_raw_machos} assumes slice lengths are unchanged and writes at
+    # their recorded ranges. Signing appends data, so every architecture's size
+    # and each following aligned offset must be recalculated.
+    # @return [void]
+    # @api private
+    def repopulate_resized_raw_machos
+      fat_arch_class = Utils.fat_magic32?(header.magic) ? Headers::FatArch : Headers::FatArch64
+      header_size = Headers::FatHeader.bytesize + (fat_archs.size * fat_arch_class.bytesize)
+      header_data = @raw_data.byteslice(0, header_size)
+      slices = +"".b
+      offset = header_size
+
+      fat_archs.zip(machos).each_with_index do |(arch, macho), index|
+        macho_offset = Utils.round(offset, 2**arch.align)
+        slices << Utils.nullpad(macho_offset - offset) << macho.serialize
+        arch_offset = Headers::FatHeader.bytesize + (index * fat_arch_class.bytesize)
+        if fat_arch_class == Headers::FatArch
+          raise FatArchOffsetOverflowError, macho_offset if macho_offset > 0xffffffff
+
+          header_data[arch_offset + 8, 8] = [macho_offset, macho.serialize.bytesize].pack("N2")
+        else
+          header_data[arch_offset + 8, 16] = [macho_offset, macho.serialize.bytesize].pack("Q>2")
+        end
+        offset = macho_offset + macho.serialize.bytesize
+      end
+
+      @raw_data = header_data + slices
+      populate_fields
+    end
+
     # Yield each Mach-O object in the file, rescuing and accumulating errors.
     # @param options [Hash]
     # @option options [Boolean] :strict (true) whether or not to fail loudly

data/lib/macho/headers.rb

@@ -60,14 +60,14 @@ module MachO
       MH_CIGAM_64 => "MH_CIGAM_64",
     }.freeze
 
-    # mask for CPUs with 64-bit architectures (when running a 64-bit ABI?)
+    # mask for 64-bit CPU architectures with 64-bit types
     # @api private
     CPU_ARCH_ABI64 = 0x01000000
 
-    # mask for CPUs with 64-bit architectures (when running a 32-bit ABI?)
+    # mask for 64-bit CPU architectures with 32-bit types (ILP32)
     # @see https://github.com/Homebrew/ruby-macho/issues/113
     # @api private
-    CPU_ARCH_ABI32 = 0x02000000
+    CPU_ARCH_ABI64_32 = 0x02000000
 
     # any CPU (unused?)
     # @api private
@@ -97,9 +97,10 @@ module MachO
     # @api private
     CPU_TYPE_ARM64 = (CPU_TYPE_ARM | CPU_ARCH_ABI64)
 
-    # 64-bit ARM compatible CPUs (running in 32-bit mode?)
+    # 64-bit ARM compatible CPUs (with 32-bit types)
     # @see https://github.com/Homebrew/ruby-macho/issues/113
-    CPU_TYPE_ARM64_32 = (CPU_TYPE_ARM | CPU_ARCH_ABI32)
+    # @api private
+    CPU_TYPE_ARM64_32 = (CPU_TYPE_ARM | CPU_ARCH_ABI64_32)
 
     # PowerPC compatible CPUs
     # @api private
@@ -450,6 +451,14 @@ module MachO
     # @api private
     MH_FILESET = 0xc
 
+    # gpu program
+    # @api private
+    MH_GPU_EXECUTE = 0xd
+
+    # gpu support functions
+    # @api private
+    MH_GPU_DYLIB = 0xe
+
     # association of filetypes to Symbol representations
     # @api private
     MH_FILETYPES = {
@@ -465,6 +474,8 @@ module MachO
       MH_DSYM => :dsym,
       MH_KEXT_BUNDLE => :kext_bundle,
       MH_FILESET => :fileset,
+      MH_GPU_EXECUTE => :gpu_execute,
+      MH_GPU_DYLIB => :gpu_dylib,
     }.freeze
 
     # association of mach header flag symbols to values
@@ -480,7 +491,7 @@ module MachO
       :MH_TWOLEVEL => 0x80,
       :MH_FORCE_FLAT => 0x100,
       :MH_NOMULTIDEFS => 0x200,
-      :MH_NOPREFIXBINDING => 0x400,
+      :MH_NOFIXPREBINDING => 0x400,
       :MH_PREBINDABLE => 0x800,
       :MH_ALLMODSBOUND => 0x1000,
       :MH_SUBSECTIONS_VIA_SYMBOLS => 0x2000,
@@ -495,9 +506,10 @@ module MachO
       :MH_DEAD_STRIPPABLE_DYLIB => 0x400000,
       :MH_HAS_TLV_DESCRIPTORS => 0x800000,
       :MH_NO_HEAP_EXECUTION => 0x1000000,
-      :MH_APP_EXTENSION_SAFE => 0x02000000,
-      :MH_NLIST_OUTOFSYNC_WITH_DYLDINFO => 0x04000000,
-      :MH_SIM_SUPPORT => 0x08000000,
+      :MH_APP_EXTENSION_SAFE => 0x2000000,
+      :MH_NLIST_OUTOFSYNC_WITH_DYLDINFO => 0x4000000,
+      :MH_SIM_SUPPORT => 0x8000000,
+      :MH_IMPLICIT_PAGEZERO => 0x10000000,
       :MH_DYLIB_IN_CACHE => 0x80000000,
     }.freeze
 

data/lib/macho/load_commands.rb

@@ -66,6 +66,10 @@ module MachO
       (LC_REQ_DYLD | 0x34) => :LC_DYLD_CHAINED_FIXUPS,
       (LC_REQ_DYLD | 0x35) => :LC_FILESET_ENTRY,
       0x36 => :LC_ATOM_INFO,
+      0x37 => :LC_FUNCTION_VARIANTS,
+      0x38 => :LC_FUNCTION_VARIANT_FIXUPS,
+      0x39 => :LC_TARGET_TRIPLE,
+      0x3a => :LC_LAZY_LOAD_DYLIB_INFO,
     }.freeze
 
     # association of symbol representations to load command constants
@@ -88,6 +92,7 @@ module MachO
       LC_ID_DYLIB
       LC_RPATH
       LC_LOAD_DYLINKER
+      LC_CODE_SIGNATURE
     ].freeze
 
     # association of load command symbols to string representations of classes
@@ -155,6 +160,10 @@ module MachO
       :LC_DYLD_CHAINED_FIXUPS => "LinkeditDataCommand",
       :LC_FILESET_ENTRY => "FilesetEntryCommand",
       :LC_ATOM_INFO => "LinkeditDataCommand",
+      :LC_FUNCTION_VARIANTS => "LinkeditDataCommand",
+      :LC_FUNCTION_VARIANT_FIXUPS => "LinkeditDataCommand",
+      :LC_TARGET_TRIPLE => "TargetTripleCommand",
+      :LC_LAZY_LOAD_DYLIB_INFO => "LinkeditDataCommand",
     }.freeze
 
     # association of segment name symbols to names
@@ -324,6 +333,8 @@ module MachO
           view = lc.view
 
           if view
+            raise LCStrMalformedError, lc if lc_str < lc.class.bytesize || lc_str >= lc.cmdsize
+
             lc_str_abs = view.offset + lc_str
             lc_end = view.offset + lc.cmdsize - 1
             raw_string = view.raw_data.slice(lc_str_abs..lc_end)
@@ -488,7 +499,7 @@ module MachO
         align = 0
         segalign = 1
 
-        while (segalign & vmaddr).zero?
+        while segalign.nobits?(vmaddr)
           segalign <<= 1
           align += 1
         end
@@ -610,7 +621,7 @@ module MachO
 
         if dylib_command.timestamp == DYLIB_USE_MARKER &&
            dylib_command.name.to_i == DylibUseCommand.bytesize
-          super(view)
+          super
         else
           dylib_command
         end
@@ -1053,11 +1064,26 @@ module MachO
       end
     end
 
+    # A load command containing the target triple used when compiling the binary.
+    # Corresponds to LC_TARGET_TRIPLE.
+    class TargetTripleCommand < LoadCommand
+      # @return [LCStr] the target triple used when compiling the binary
+      field :triple, :lcstr, :to_s => true
+
+      # @return [Hash] a hash representation of this {TargetTripleCommand}
+      def to_h
+        {
+          "triple" => triple.to_h,
+        }.merge super
+      end
+    end
+
     # A load command representing the offsets and sizes of a blob of data in
     # the __LINKEDIT segment. Corresponds to LC_CODE_SIGNATURE,
     # LC_SEGMENT_SPLIT_INFO, LC_FUNCTION_STARTS, LC_DATA_IN_CODE,
     # LC_DYLIB_CODE_SIGN_DRS, LC_LINKER_OPTIMIZATION_HINT, LC_DYLD_EXPORTS_TRIE,
-    # LC_DYLD_CHAINED_FIXUPS, or LC_ATOM_INFO.
+    # LC_DYLD_CHAINED_FIXUPS, LC_ATOM_INFO, LC_FUNCTION_VARIANTS,
+    # LC_FUNCTION_VARIANT_FIXUPS, or LC_LAZY_LOAD_DYLIB_INFO.
     class LinkeditDataCommand < LoadCommand
       # @return [Integer] offset to the data in the __LINKEDIT segment
       field :dataoff, :uint32
@@ -1065,6 +1091,25 @@ module MachO
       # @return [Integer] size of the data in the __LINKEDIT segment
       field :datasize, :uint32
 
+      # @param context [SerializationContext] the context
+      # @return [String] the serialized fields of the load command
+      # @api private
+      def serialize(context)
+        raise LoadCommandNotSerializableError, type unless serializable?
+
+        format = Utils.specialize_format(self.class.format, context.endianness)
+        [cmd, self.class.bytesize, dataoff, datasize].pack(format)
+      end
+
+      # The embedded signature referenced by this command.
+      # @return [CodeSigning::SuperBlob]
+      # @raise [CodeSigningError] if this is not an LC_CODE_SIGNATURE command
+      def superblob
+        raise CodeSigningError, "#{type} does not contain a code signature" unless type == :LC_CODE_SIGNATURE
+
+        CodeSigning::SuperBlob.new(view.raw_data.byteslice(dataoff, datasize))
+      end
+
       # @return [Hash] a hash representation of this {LinkeditDataCommand}
       def to_h
         {